smihael/Combining several imperfect classifiers to reduce overall classification error.ipynb

## Combining several imperfect classifiers to reduce overall classification error.ipynb
{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "66a3c05e",
   "metadata": {},
   "source": [
    "## Combining several imperfect classifiers to reduce overall classification error\n",
    "\n",
    "This notebook illustrates one of the basic concepts in ensemble learning.\n",
    "\n",
    "By combining classifiers that are trained on different subsets of the training data, it is possible to acheive superior classifier performance.\n",
    "\n",
    "Figure is adapted from an example published in:\n",
    "- Robi Polikar. Ensemble based systems in decision making. Circuits and Systems Magazine, IEEE, 6(3):21–45, 2006.\n",
    "  Link to the original publication: https://doi.org/10.1109/MCAS.2006.1688199\n",
    "  \n",
    "\n",
    "Code author: [smihael](https://github.com/smihael)\n",
    "\n",
    "Licence: CC-BY-SA 4.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "34505114",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn import datasets\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.ensemble import VotingClassifier\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.colors import ListedColormap\n",
    "\n",
    "from matplotlib.patches import Patch\n",
    "from matplotlib.lines import Line2D"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "82d86921",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a meshgrid for plotting\n",
    "def create_meshgrid(X, h=.02):\n",
    "    x_min, x_max = X[:, 0].min() - 1, X[:, 0].max() + 1\n",
    "    y_min, y_max = X[:, 1].min() - 1, X[:, 1].max() + 1\n",
    "    xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))\n",
    "    return xx, yy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "13ca2a30",
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_decision_boundaries(clf, X, y, xx, yy, Z, title, subplot, \n",
    "                             X_sub=None,\n",
    "                             color='lightgray',\n",
    "                             marker_types=['^', 's', 'o'],\n",
    "                             fill_areas=False,\n",
    "                             plot_boundary=True,\n",
    "                             show_legend=False):\n",
    "    \n",
    "    if fill_areas:\n",
    "        subplot.contourf(xx, yy, Z, alpha=0.1, cmap=ListedColormap(['lightgray','gray','darkgray']))\n",
    "    elif plot_boundary:\n",
    "        subplot.contour(xx, yy, Z, alpha=0.5, cmap=ListedColormap([color]))\n",
    "\n",
    "    # Plot all points with gray color first\n",
    "    for idx, cl in enumerate(np.unique(y)):\n",
    "        subplot.scatter(x=X[y == cl, 0], y=X[y == cl, 1], c='0.5', marker=marker_types[idx],\n",
    "                        label=f'Class {cl}', edgecolors='k', s=20, alpha=0.5)\n",
    "\n",
    "    # Plot points from the training subset with specified color\n",
    "    if X_sub is not None:\n",
    "        for idx, cl in enumerate(np.unique(y)):\n",
    "            indices = np.where(np.logical_and(y == cl, np.isin(range(len(X)), X_sub)))[0]\n",
    "            if len(indices) > 0:\n",
    "                subplot.scatter(x=X[indices, 0], y=X[indices, 1], c=color, marker=marker_types[idx],\n",
    "                                label=f'Class {cl} (subset)', edgecolors='k', s=20, alpha=1)\n",
    "\n",
    "    subplot.set_xlim(xx.min(), xx.max())\n",
    "    subplot.set_ylim(yy.min(), yy.max())\n",
    "    subplot.set_xlabel('Feature 1')\n",
    "    subplot.set_ylabel('Feature 2')\n",
    "    subplot.set_title(title)\n",
    "    if show_legend:\n",
    "        subplot.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "b3113ea0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a synthetic dataset\n",
    "# X, y = datasets.make_classification(n_samples=100, n_features=2, n_redundant=0, n_clusters_per_class=1, n_classes=3, random_state=5)\n",
    "\n",
    "# ... or use iris dataset\n",
    "iris = datasets.load_iris()\n",
    "X = iris.data[:, [0, 2]]\n",
    "y = iris.target\n",
    "\n",
    "# Split the data into three different subsets\n",
    "np.random.seed(2)\n",
    "subset_indices_1 = np.random.choice(range(len(X)), size=int(len(X)/3), replace=False)\n",
    "subset_indices_2 = np.random.choice(list(set(range(len(X))) - set(subset_indices_1)), size=int(len(X)/3), replace=False)\n",
    "subset_indices_3 = list(set(range(len(X))) - set(subset_indices_1) - set(subset_indices_2))\n",
    "\n",
    "X_sub1, y_sub1 = X[subset_indices_1], y[subset_indices_1]\n",
    "X_sub2, y_sub2 = X[subset_indices_2], y[subset_indices_2]\n",
    "X_sub3, y_sub3 = X[subset_indices_3], y[subset_indices_3]\n",
    "\n",
    "# Define the base classifiers, trained on different subsets\n",
    "clf=[None]*3\n",
    "params={'kernel': 'rbf','random_state': 0}\n",
    "clf[0] = SVC(**params).fit(X_sub1, y_sub1)\n",
    "clf[1] = SVC(**params).fit(X_sub2, y_sub2)\n",
    "clf[2] = SVC(**params).fit(X_sub3, y_sub3)\n",
    "\n",
    "# Define an ensemble of the three classifiers\n",
    "ensemble = VotingClassifier(estimators=[('a', clf[0]), ('b', clf[1]), ('c', clf[2])], voting='hard')\n",
    "ensemble.fit(X, y)\n",
    "\n",
    "# Ceate meshgrid\n",
    "xx, yy = create_meshgrid(X)\n",
    "\n",
    "# Get decision boundaries   \n",
    "Z=[None]*3\n",
    "for i in range(3):\n",
    "    Z[i] = clf[i].predict(np.c_[xx.ravel(), yy.ravel()])\n",
    "    Z[i] = Z[i].reshape(xx.shape)\n",
    "    \n",
    "    \n",
    "Z_ensamble = ensemble.predict(np.c_[xx.ravel(), yy.ravel()])\n",
    "Z_ensamble = Z_ensamble.reshape(xx.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "a297c31c",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting decision boundaries\n",
    "fig, axs = plt.subplots(2, 3, figsize=(15, 10), gridspec_kw={'height_ratios': [1, 1], 'hspace': 0.75})\n",
    "\n",
    "# Define colors for different subsets\n",
    "colors = ['blue', 'red', 'green']\n",
    "\n",
    "plot_decision_boundaries(clf[0], X, y, xx, yy, Z[0], 'Classifier 1', axs[0, 0], subset_indices_1, color=colors[0])\n",
    "plot_decision_boundaries(clf[1], X, y, xx, yy, Z[1], 'Classifier 2', axs[0, 1], subset_indices_2, color=colors[1])\n",
    "plot_decision_boundaries(clf[2], X, y, xx, yy, Z[2], 'Classifier 3', axs[0, 2], subset_indices_3, color=colors[2],fill_areas=True)\n",
    "\n",
    "# Plot all decision boundaries together\n",
    "plot_decision_boundaries(ensemble, X, y, xx, yy, Z_ensamble, 'Different decision boundaries', axs[1, 0], plot_boundary=False)\n",
    "d=[None]*3\n",
    "for i in range(3):\n",
    "    d[i] = axs[1, 0].contour(xx, yy, Z[i], alpha=0.1, colors=colors[i])\n",
    "    #axs[1, 0].clabel(d[i], fmt=f'Classifier {i}', inline_spacing=10, fontsize=8)\n",
    "\n",
    "\n",
    "# Plot ensamble\n",
    "plot_decision_boundaries(ensemble, X, y, xx, yy, Z_ensamble, 'Ensemble and\\ncombined decision boundary', axs[1, 1])\n",
    "\n",
    "\n",
    "# Clear the last subplot for the legend\n",
    "axs[1, 2].clear()\n",
    "axs[1, 2].axis('off')\n",
    "\n",
    "# Custom legend for the entire figure\n",
    "legend_elements = [\n",
    "    Line2D([0], [0], color='w', markerfacecolor='lightgray', markeredgecolor='k', marker='^', markersize=10, label='Class 0'),\n",
    "    Line2D([0], [0], color='w', markerfacecolor='lightgray', markeredgecolor='k', marker='s', markersize=10, label='Class 1'),\n",
    "    Line2D([0], [0], color='w', markerfacecolor='lightgray', markeredgecolor='k', marker='o', markersize=10, label='Class 2'),\n",
    "    Line2D([0], [0], color='w', markerfacecolor=colors[0], marker='*', markersize=10, label='Classifier 1\\ntraining subset'),        \n",
    "    Line2D([0], [0], color='w', markerfacecolor=colors[1], marker='*', markersize=10, label='Classifier 1\\ntraining subset'),        \n",
    "    Line2D([0], [0], color='w', markerfacecolor=colors[2], marker='*', markersize=10, label='Classifier 1\\ntraining subset'),        \n",
    "\n",
    "    Line2D([0], [0], color=colors[0], lw=2, label='Classifier 1\\nboundary'),\n",
    "    Line2D([0], [0], color=colors[1], lw=2, label='Classifier 2\\nboundary'),\n",
    "    Line2D([0], [0], color=colors[2], lw=2, label='Classifier 3\\nboundary'),\n",
    "]\n",
    "axs[1, 2].legend(handles=legend_elements, loc='right')\n",
    "\n",
    "\n",
    "# The second subplot in the second row should be positioned to make space for the horizontal arrow\n",
    "axs[1, 1].set_position([0.5, axs[1,0].get_position().y0,\n",
    "                        axs[0, 0].get_position().width,\n",
    "                        axs[0, 0].get_position().height])\n",
    "\n",
    "\n",
    "# Add arrows from the top row to the 4th subplot, shortening them so the heads are not overlapped (patchB)\n",
    "\n",
    "arrow_end_x = axs[1, 0].get_position().x0 + 0.5 * axs[1, 0].get_position().width\n",
    "arrow_end_y = axs[1, 0].get_position().y0 + 0.5 * axs[1, 0].get_position().height\n",
    "\n",
    "for i in range(3):\n",
    "    axs[0, i].annotate(\"\", xy=(arrow_end_x, arrow_end_y), xycoords='figure fraction', \n",
    "                        xytext=(0.5, -0.2), textcoords='axes fraction',\n",
    "                        arrowprops=dict(arrowstyle=\"simple\",\n",
    "                            color=colors[i],\n",
    "                            patchB=axs[1,0],\n",
    "                            shrinkB=20))\n",
    "\n",
    "    \n",
    "# Add a big horizontal arrow between the subplots in the second row\n",
    "\n",
    "arrow_start_x = axs[1, 0].get_position().x1  # x position where the fourth subplot ends\n",
    "arrow_end_x = axs[1, 1].get_position().x0    # x position where the fifth subplot starts\n",
    "\n",
    "arrow_y = axs[1, 0].get_position().y0 + axs[1, 0].get_position().height/3\n",
    "\n",
    "ma=axs[1, 0].annotate(\"\", xy=(arrow_end_x-0.14, arrow_y), xycoords='figure fraction', \n",
    "                    xytext=(arrow_start_x-0.07, arrow_y), textcoords='figure fraction',\n",
    "                    arrowprops=dict(arrowstyle=\"simple,head_width=0.8,head_length=0.8\", color=\"0.5\", lw=5))\n",
    "\n",
    "# Let's display the plot to see the current output\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "06e37e04",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classifier 1 accuracy: 0.93\n",
      "Classifier 2 accuracy: 0.93\n",
      "Classifier 3 accuracy: 0.95\n",
      "Ensemble classifier accuracy: 0.96\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "# We won't do the classical train/test split with train_test_split(X, y, test_size=0.3), \n",
    "# as this is specific example, where we want to see graphically\n",
    "# what training data went in each of the base classifiers and\n",
    "# how the classifier performs on the overall dataset. \n",
    "\n",
    "X_test = X\n",
    "y_test = y\n",
    "\n",
    "accuracies = []\n",
    "\n",
    "for i, classifier in enumerate(clf):\n",
    "    y_pred = classifier.predict(X_test)\n",
    "    accuracy = accuracy_score(y_test, y_pred)\n",
    "    accuracies.append(accuracy)\n",
    "    print(f\"Classifier {i+1} accuracy: {accuracy:.2f}\")\n",
    "\n",
    "# Print accuracy of the ensemble classifier\n",
    "y_pred_ensemble = ensemble.predict(X_test)\n",
    "accuracy_ensemble = accuracy_score(y_test, y_pred_ensemble)\n",
    "print(f\"Ensemble classifier accuracy: {accuracy_ensemble:.2f}\")\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}

## grid_search.py
from sklearn.svm import SVC
from sklearn.ensemble import VotingClassifier
from sklearn.metrics import accuracy_score
from sklearn.model_selection import ParameterGrid


# Define the search space
subset_seed_range = range(0, 100)  # Range of seeds to try
param_grid = {
    'C': [0.1, 1, 10],
    'gamma': [0.001, 0.01, 0.1, 1],
    'kernel': ['rbf'],
    'random_state': range(0,5)
}
all_combinations = [(seed, params) for seed in subset_seed_range for params in ParameterGrid(param_grid)]

# Initialize variables to track the best performance
best_ensemble_accuracy = 0
best_combination = None

n_passed = 0

# Perform the search
for seed, params in all_combinations:
    #print(seed)
    #print(params)
    # Create the subsets using the seed
    np.random.seed(seed)
    subset_indices_1 = np.random.choice(range(len(X)), size=int(len(X)/3), replace=False)
    subset_indices_2 = np.random.choice(list(set(range(len(X))) - set(subset_indices_1)), size=int(len(X)/3), replace=False)
    subset_indices_3 = list(set(range(len(X))) - set(subset_indices_1) - set(subset_indices_2))

    X_sub1, y_sub1 = X[subset_indices_1], y[subset_indices_1]
    X_sub2, y_sub2 = X[subset_indices_2], y[subset_indices_2]
    X_sub3, y_sub3 = X[subset_indices_3], y[subset_indices_3]

    # Train SVC models on each subset with the given parameters
    classifiers = []
    for i, (X_sub, y_sub) in enumerate([(X_sub1, y_sub1), (X_sub2, y_sub2), (X_sub3, y_sub3)]):
        clf = SVC(**params).fit(X_sub, y_sub)
        classifiers.append(clf)

    # Create the ensemble
    ensemble = VotingClassifier(estimators=[(f'clf{i}', clf) for i, clf in enumerate(classifiers)], voting='hard')
    ensemble.fit(X, y)

    # Evaluate the ensemble and base classifiers
    ensemble_accuracy = accuracy_score(y, ensemble.predict(X))
    base_accuracies = [accuracy_score(y, clf.predict(X)) for clf in classifiers]

    # We don't want 'bad' examples
    passed = all(acc < ensemble_accuracy for acc in base_accuracies)

    if passed:
        n_passed = n_passed + 1

    # Check if this combination is the best so far
    if ensemble_accuracy > best_ensemble_accuracy and passed:
        best_ensemble_accuracy = ensemble_accuracy
        best_combination = (seed, params)

    #print("----")

# Output the results
print(f"Best ensemble accuracy: {best_ensemble_accuracy:.2f}")
print(f"Best combination: Seed={best_combination[0]}, Parameters={best_combination[1]}")

## Združevanje_več_klasifikatorjev (alt).ipynb

      
Display the source blob

    
Display the rendered blob

    
    Raw
  

              Združevanje_več_klasifikatorjev (alt).ipynb
            
          
        Loading

      Sorry, something went wrong. Reload?
      Sorry, we cannot display this file.
      Sorry, this file is invalid so it cannot be displayed.
      
          Viewer requires iframe.
      
    
## Združevanje_več_klasifikatorjev.ipynb

      
Display the source blob

    
Display the rendered blob

    
    Raw
  

              Združevanje_več_klasifikatorjev.ipynb
            
          
        Loading

      Sorry, something went wrong. Reload?
      Sorry, we cannot display this file.
      Sorry, this file is invalid so it cannot be displayed.
      
          Viewer requires iframe.
	from sklearn.svm import SVC
	from sklearn.ensemble import VotingClassifier
	from sklearn.metrics import accuracy_score
	from sklearn.model_selection import ParameterGrid


	# Define the search space
	subset_seed_range = range(0, 100) # Range of seeds to try
	param_grid = {
	'C': [0.1, 1, 10],
	'gamma': [0.001, 0.01, 0.1, 1],
	'kernel': ['rbf'],
	'random_state': range(0,5)
	}
	all_combinations = [(seed, params) for seed in subset_seed_range for params in ParameterGrid(param_grid)]

	# Initialize variables to track the best performance
	best_ensemble_accuracy = 0
	best_combination = None

	n_passed = 0

	# Perform the search
	for seed, params in all_combinations:
	#print(seed)
	#print(params)
	# Create the subsets using the seed
	np.random.seed(seed)
	subset_indices_1 = np.random.choice(range(len(X)), size=int(len(X)/3), replace=False)
	subset_indices_2 = np.random.choice(list(set(range(len(X))) - set(subset_indices_1)), size=int(len(X)/3), replace=False)
	subset_indices_3 = list(set(range(len(X))) - set(subset_indices_1) - set(subset_indices_2))

	X_sub1, y_sub1 = X[subset_indices_1], y[subset_indices_1]
	X_sub2, y_sub2 = X[subset_indices_2], y[subset_indices_2]
	X_sub3, y_sub3 = X[subset_indices_3], y[subset_indices_3]

	# Train SVC models on each subset with the given parameters
	classifiers = []
	for i, (X_sub, y_sub) in enumerate([(X_sub1, y_sub1), (X_sub2, y_sub2), (X_sub3, y_sub3)]):
	clf = SVC(**params).fit(X_sub, y_sub)
	classifiers.append(clf)

	# Create the ensemble
	ensemble = VotingClassifier(estimators=[(f'clf{i}', clf) for i, clf in enumerate(classifiers)], voting='hard')
	ensemble.fit(X, y)

	# Evaluate the ensemble and base classifiers
	ensemble_accuracy = accuracy_score(y, ensemble.predict(X))
	base_accuracies = [accuracy_score(y, clf.predict(X)) for clf in classifiers]

	# We don't want 'bad' examples
	passed = all(acc < ensemble_accuracy for acc in base_accuracies)

	if passed:
	n_passed = n_passed + 1

	# Check if this combination is the best so far
	if ensemble_accuracy > best_ensemble_accuracy and passed:
	best_ensemble_accuracy = ensemble_accuracy
	best_combination = (seed, params)

	#print("----")

	# Output the results
	print(f"Best ensemble accuracy: {best_ensemble_accuracy:.2f}")
	print(f"Best combination: Seed={best_combination[0]}, Parameters={best_combination[1]}")